15 February 2008

Performance Improvement, part three

I've written about the operational crisis our ER built up to over a series of several years, and the general themes of the turnaround plan we developed. Today I'd like to break it down into a little more granular level of detail, and address the "process analysis" and improvements we developed. It has been said that in the average ER seeing 20-30,000 patients annually, you can just let things happen and, provided adequate staffing, the patient flow will remain adequate. But patient flow is analagous to fluid dynamics, wherein the resistance to flow increases exponentially with flow rates: as an ER scales up, the friction increases disproportionately to the point that process break down and stop, unless they have been carefully engineered.

Our ED was using the same old ad hoc processes we'd had for a decade or more. We had a white board to keep track of the patients and order status, paper charts, paper orders, and a free-floating staffing model. Stone age.

The first thing you do when re-engineering the core processes is to flowchart it: in its simplest form, this exercise consists of putting each essential step on a post-it note and laying them out in the sequence they occur. You identify the rate-limiting steps, and try to identify inefficiencies, such as tasks done in serial which could be done in parallel, or a low-value but time-consuming step being performed before a high-value step.

One example is registration. It's crucial and necessary, of course, but adds nothing to the visit from the standpoint of patient care. Yet for some reason registration was traditionally performed before the physician's assessment, sometimes even before triage! For most patients, there is a lot of "down time" while the assessment is underway when the registration can be performed. So one new process was to do a "quick reg" consisting of name & DOB on patient arrival, just to get them in the computer, and later, the registrar would come back to the room with a portable computer to complete the data acquisition. There was a lot of cultural resistance to this, but it removed a key choke point.

For us, the limiting factor over-all was and is the size of our ER. We have 50 beds, but that may not be the best way to define the size of a facility. I think of it more in the sense of an ER's "Virtual size." Given that there are, for each bed, 60*24 = 1440 bed-minutes per day, the functional capacity of our ER was 50*1440 = 72,000 bed-minutes. At an average Length of Stay (LOS) of 260 minutes, the operational capacity of the unit was 276 patients/day. Our annual census of 100,000 results in average volume of 273 patients/day! We were operating exactly at our full capacity, in fact over, since patients present in surges depending on time of day and day of the week. We "created" extra beds by sticking patients in the hallways, and leaving patients in the waiting room until the daily surge tapered off after midnight.

Given that new beds were not in the budget, the challenge became this: how can we reduce LOS and by doing so free up some operational capacity and increase the "virtual size" of the ER?

Throughput. It's all about throughput efficiency. This has been the major challenge of my administrative career, and is why this blog is titled "Movin' meat," since the topic is always on my mind.

So: attack the bottle necks and wasted time.

Triage. A good triage takes maybe ten minutes, including documentation of the irrelevant information the Joint Commission requires us to collect. During peak hours, 10-15 patients present through the front doors. One triage nurse cannot handle that sort of volume. So we added staff in triage, flexing up and down, to triage 2-3 patients simultaneously.

Waiting room time. When patients are in the waiting room for an extended time, it makes sense to initiate the work-up from triage. This can be done effectively through the use of standing orders. At a higher cost, more complete work-ups can be initiated by staffing a midlevel provider such as a PA or even a physician at triage. Some fraction of patients can even be discharged directly from triage. It greatly increases patient satisfaction. However, the payroll cost of this measure is generally not offset by any increase in patient fees, and care can become somewhat fragmented when different providers begin and conclude the work-up.

Initiating a standard order set, including labs and x-rays, while patients are in the waiting room can speed things up very effectively, since by the time the patient is placed in a bed there may well be results on the chart, expediting the disposition. This requires additional resources (phlebotomy, e.g.) at triage, further reducing resources in the main ED. The most time-consuming tests, like CT scans, are generally not included in triage protocols, for good reason, which places an upper limit on how much utility can be extracted from waiting room time. There is also often provider resistance to standard order sets; each doc likes to do things his or her own way and in some cases can sabotage the protocols by complaining and decreasing nursing compliance.

All of these interventions lose value dramatically if and when the patients stop spending significant time in the waiting room. Ultimately, the ability to utilize the waiting room time is limited by the absence of a attending provider to personalize and direct the care. When there are unavoidable waits it makes sense to use the time as best you can, but for maximum efficiency it is best to find a way to eliminate the wait and bed patients immediately.

Bed-to-physician. Wasted time; worse than "waiting room time" because the patient is wasting valuable bed-minutes. I'll discuss closing this time gap later.

Order entry. Tends to be idiosyncratic to each institution. In an ideal world, one-click physician order entry would be ideal. Technical challenges preclude that for us, and we remain with a kludge-y paper-based workaround. But the time between the entry of the order and its implementation can be significant if not well-designed.

Lab and X-ray. Creation of dedicated lab tech positions in the ER to enter the orders, draw the blood, and act as gatekeeper of results resulted in quicker turn-around time (TAT) for lab results. The use of point-of-care testing such as iStat, iStat troponin, urinalysis, and blood gasses in the ER can greatly decrease the TAT until you get the most critical lab data resulted. However, this requires buy-in from the pathologists, often skeptical of these devices, and dedicated space in the ER. It's a thing of beauty to see a patient with chest pain, have an ECG and Troponin in hand, and be ready for dispostion in ten minutes. For us, x-ray was not a significant delay. Digital radiology with 24/7 wet-reads by a large local group, communicated by fax/phone prn, left little room for improvement on this front. Your mileage may vary.

Disposition to departure. This falls into two main subgroups depending on the patient's destination: home or an inpatient bed. Patients going home are typically not too much of a challenge, though some patients with social challenges (i.e. need a ride, still drunk, etc) can have an extended epilogue to their ED work-up. Making this a focus and having pro-active nurses to prompt the doc to discharge the patient can help, but it's not usually a make-or-break point. We have experimented with having discharge planners in the ED. It can help, but honestly, they tend to spend most of their productive energy upstairs on the wards.

Admitted patients, however, are a huge problem. Admit-to-bed times can be 120 minutes or more, and in an ER with 30+ admissions daily, you can see how that cannibalizes the availability of ER beds for new patients.

I'll leave aside for the moment any discussion of "patient boarding," which is a much bigger and more intractable issue. When there are no inpatient beds, or no ICU beds, patients will just reside in ERs nationwide, sometimes for days. The solution to that is more beds, more staff, more hospitals. Not forthcoming any time soon. I'll limit this discussion to the "normal" hospital which has inpatient beds but it typically slow in finding them.

Institutional Commitment
The most important and difficult step to solving this is getting buy-in from the rest of the hospital. Modern American healthcare institutions tend to devolve into a bunch of petty feudal fiefdoms with very limited interests beyond the end of the hallway. The inpatient wards tended to view ER overcrowding as an ER problem. They didn't see their units as having a role in the cause or the solution to ER congestion. That is a mindset which is difficult to change and requires hospital leadership to take charge and make ER congestion a hospital-wide issue.

There must be centralized authority in one person for bed assignment and turn-over. Historically this resided in some sort of "head nurse" or "Nurse Supervisor" who bear a host of administrative responsibilities. But in a large facility, this is a big enough job to dedicate one person to. We euphemistically call this person the "Bed czar," and he or she has power to determine bed placement, shuffle patients around, overflow into other units, and act as a gadfly to inpatient nurses and ancillary services to be ready for incoming patients. Medical staff need to be prompted to make "discharge rounds" the first daily priority, and housekeeping needs to be organized into rapid response teams to turn over vacated rooms immediately. Major hotels do this better than anybody, and the medical industry can learn from their processes.

Other barriers to overcome include the tradition of "calling report," which takes forever. The accepting nurse is busy, or on break, or "can you call back in 40 minutes because it's change of shift and we're giving report?" But it can be replaced with a faxed report, or bedside report, or designating an inpatient nurse as the "admitting nurse" for all ED admits. And maybe the bed is ready but there are no physician orders on the chart. Or the admitting doc wants to keep the patient in the ER so they can come ans see them there. On and on and on. In our ED we have a RN dedicated to being the "admissions facilitator" whose entire job is to keep tabs on all the patients designated for admission and knock down the obstacles to getting them upstairs.

Most controversial is a process of boarding hallway patients in inpatient hallways. This is a new concept, endorsed by the Joint Commission as a strategy to ease ED overcrowding.
It's likely to incite civil war wherever implemented, but the theory is sound -- an admitted patient is less vulnerable in an inpatient hall bed than in a chaotic ED hall bed, and nothing motivates the inpatient team to "find" a bed than the threat of a hallway patient. We've not had to implement this plan, but it exists, and in any crisis remains our ace in the hole to decompress the ED.

I'll leave off there for the moment, though so much more could be said on these few topics. Already I grow long-winded. Tomorrow I will write about the new management philosophy that the ED team developed, and when I say "tomorrow" I probably mean "in a week or so" since it turns out that these posts are time-intensive to write, and I may, just may, choose to spend some time with my kids this weekend instead.

Indulge me on this....


  1. Thanks for this! This has been a great series.

    We already have lots of the things you've suggested in place, and we just added a "quick reg" for our ambulatory patients (previously only ambulance arrivals would be quick-reg'd). Now all acute and sub-acute patients are registered directly after triage, but the fast-track patients wait.

    I wish that we had order sets that the nurses and the triage nurses could implement (since I draw all bloods that I anticipate will be ordered as soon as the patient's butt hits the stretcher, but they often sit around for an hour or more until the patient is seen), but administration is opposed to that. They say that since this is a teaching hospital, the residents need to learn what to order. Also, nursing administration is fearful that an already overburdened nursing staff would get blamed for "missing something" or some other reason, and therefore don't want nurses doing the ordering.

  2. Hello,

    Nice post. The term "flowchart" caught my eye and I wanted to let you know about easy software for making flowcharts to improve healthcare processes. Its called the SmartDraw Healthcare Edition. You can download a free trial here: www.smartdrawhealth.com. In addition to flowcharts, you can create schedules, facility floor plans, Gantt charts, timelines,org charts, and lots other visual tools to get organized. Hope this helps!

  3. i hate the joint commission

  4. I am so happy to hear i-STAT is impacting your ED in a positive way to improve pateint care. I hear they are looking for doctor's for the speaker's bureau to present to other clinicians.

  5. Fantastic ideas on finding waste in process flow. I agree that the variability is what can kill the process so we are trying hard to work on creating pull systems in our clinic. Rather than "stacking" patients in open rooms so the provider doesn't waste any time we're trying to have keep one or two patients ahead. It helps eleminate bottlenecks further in the system. We create the schedule and ask everyone to stick to it rigorously. www.waittimes.blogspot.com

  6. Another big problem in the 50 bed ER I work at that bogs us down is unnecessary tests and treatments. Prime example-healthy adult in their 20s with a complaint of vomiting "all day" who has no abd pain, no ketones in their urine. Looks fine. Instead of giving them a pat on their head and an rx for zofran or phenergan we spend 2 hours and god knows how many hundreds of dollars starting an IV and giving IV meds and fluids. Big waste of time, ER bed and the nurses time, who could be expediting someone else's care.

    Or the hero docs who think they're going to be the one who is suddenly going to figure out why the patient who has been seen for functional abd pain weekly for the past 4 years is going to find out what's causing it. Oh, it just might be something interesting. Yea, right. I'm sure. So we do the 6 hour work up for the 25th time and send them home with a diagnosis of "abd pain."

    Also at any given time we have anywhere from 4 to 14 of our 50 beds occupied by psych patients waiting for an inpatient mental health bed to become available for admission. Some psych patients wait up to 72 hours for a bed and that is not an exaggeration. It is quite common with probably 36 hours being the average wait for an inpatient psych bed in the Portland area.

    When we've got 1/4 of our beds bogged down with psych patients who aren't going anywhere you can be damned sure that the wait in the lobby goes up, sometimes to 4 hours. Kidney stones might end up waiting an hour or two. Here's a bag to vomit in, find a nice seat to writhe around in.


Note: Only a member of this blog may post a comment.